Romanov

Algorithms of auto-correction in documents were analyzed. The authors proposed a method of correcting grammar mistakes by means of a morphological analysis. The authors present a bundled software allowing to recognize text images by means of neural network and to correct mistakes with the usage of the proposed method. Creation of a special storage for the dictionary allowed to increase productivity and decrease the complexity of word selection for the operation of the bundled software. The dependence of correction accuracy on the mistake location in the word was identified.